Has anyone figured out a workflow to generate a 3D model of a part automatically, by scanning the part with a turntable, or multiple-camera system? I have a turntable I built years ago that could be computer-controlled (right now it will rotate a programmable number of degrees each time a button is pressed). I have seen some impressive demos using iOS and the new iPhones to generate 3D models from photographic input, but the file format is USDZ, and I suspect that it won’t convert to STEP very well.
I don’t have much of a need for 3D PCB models, but if it was easy enough, I would take the time to make them. I doubt I’m ever going to want to draw them by hand.
Well as long as we’re speculating, how about a system that could go from dimensioned drawings to 3D models? Advantage is you wouldn’t need to have the object to hand.
I suspect that’s easily doable.
I have a cheapo CAD program (TurboCAD) that can do that. One still has to draw the part, though - even if it’s just copying the dimensions on the data sheet.
The general term for reconstructing a 3D object from 3D photographs is called Photogrammetry
But usually it does not create very nice 3D models, but a quite chaotic mesh with lots of vectors.
There are also systems that use a rotating table, a laser line pointer and a camera. I did a quick search and these things seem to have evolved. I did not look very far, but quickly found this one:
And another site with a list of those things:
But this is all terribly off-topic for this KiCad forum, which probably means this topic will be locked soon…
Well I was hoping for a more automatic system, one you could feed the PDF of the datasheet and it spits out a STEP file after a while. It would have to have smarts, e.g. to recognise engineering drawing conventions, like hidden edges, etc. Might need some manual assistance in case there are several sets of drawings in the datasheet.
Incidentally how is your collection of ogs coming along? (Joke about your handle. )
Many folks with 3D-printers, including myself, have done it.
I made a ‘homemade’ scanner from YouTube tutorial and these days you can buy add-on’s and 3D-printers that come with them. Here’s a page/site full of them…
Results range from ‘Questionable to Good-enough’. But, in the end, the real question is ‘For what purpose’ does a Kicad-user need Step (or wrl)?
Kicad uses them for only one purpose; the Graphic rendering in 3D-viewer. Sure, Kicad can export a Step of an existing Step. So what? User can use the resulting STL from the scanner to export Step from CAD software and some 3D-print software. Again, ‘Why’?
With or, without that exporting and/or using the KSU/StepUp plugin in FreeCad, a user still needs the CAD skills to do something with it.
That suggests (to me), a user may as well focus on the couple of hours it will take to learn some Fundamental CAD skills - then, a user will have a real-useful tool in the tool-chain.
I’ve been coming up to speed on Fusion360 for about a year now. I’m at a point where I can make acceptable models for many components where the manufacture doesn’t provide one. When selecting parts, I give a high priority to any vendor that has step files. I’m doing a project now where it took about 4 hours to make models for 4 different widgets (HDMI switch, HDMI splitter, data casting box, wall wort) so I can work to place them and make a base-plate that can be sent out for manufacturing. The ability to also use the 3D modeling tool for sheet metal give you more options to do projects.
I see parallels here to a desire for autorouting. Everyone wants an “easy” button. If you know your tools (KiCad in this case), then you really don’t need autorouting. You need good component placement skills. The idea that you can get or make a scanner, and then run it, and have perfect 3D models seems unlikely. When you consider the time to get all that set up, and then the clean-up you’ll have to do in some 3D design package, it’s highly likely that you would be better off spending “coming up to speed time on the scanner” by learning a 3D design program so you can make models. And then you can also make 3D print models for yourself. And make sheet metal designs to send out for fabrication. And chassis for your circuit boards.
The models I make for parts are all “bricks on bricks”. I just want mounting holes and space occupied. While it’s great fun to get a detailed model, such as you can get from CUI and other connector makers, you really just want a way to be sure everything fits.
I’m not saying the time invested in getting a scanner set up is a waste of time, I’m saying that one should consider the time it takes to learn and get it all working, and then weigh that against getting skill with a 3D design package where you can not only make models for things you have, but design parts for which theree is no exiting part. The ability to export a step file in KiCad makes designing a chassis for a PCB easy. My first efforts a year ago took about 20 hours, with about half that time screaming at the screen. (I always new mechanical engineers were “different”, that belief is confirmed!) The last PCB I did I did the chassis only took an evening and it was a perfect fit.
It’s often a myth that coming up to speed on a new tool will increase productivity. The whole time you are learning (or building) that tool, you could be doing other things or learning more versatile tools. I believe that KiCad and 3D layout programs are integral to each other and both are needed to built the things we want to build. PCBs need chassis, and chassis need PCBs. So learn them both.
Gratuitous picture to make a point here:
Here is the 3D viewer in KiCad. And here is the design for the chassis, which uses the step file that KiCad made. I had to make the models for the RCA phono jacks, the transformers (yellow squares), the panel LEDs and the screw terminal components. U6 (upper right) didn’t have a model for the package, but it’s clear the chip will fit the footprint I made for it.
)
